Planning in a claims environment
Claims is a different planning problem
Workforce planning frameworks were built for service and sales contact centres — predictable arrival patterns, voice-led work, single-call resolution, well-behaved Erlang maths. Drop those frameworks into a claims environment and they break in interesting ways. Claims volumes are partly random (weather, fires, large losses). Claims work mixes a short-handling FNOL function with multi-week or multi-month case handling. Claims roles are licensed, regulated, and structurally constrained in who can do what. Claims has a regulator paying attention. And on a bad day, claims volume jumps tenfold inside 12 hours and the operation has to be ready.
This article walks through what makes claims different, how to forecast it sensibly, how to staff it, and how to manage the surge events that define the discipline.
The two halves of claims operations
Most insurance claims operations have two operational halves and the planning model has to handle both.
FNOL (First Notification of Loss). The inbound channel where the customer tells you something has happened. Typically voice-led, often with digital channels alongside. Behaves more like a normal contact centre — intraday arrival pattern, AHT, SLA in seconds. Forecastable in 15-minute intervals.
Claims handling. The case-management work after FNOL. The handler investigates, validates, settles, and closes the claim over days, weeks, or months. The unit of work isn’t a call — it’s a case. The right capacity metric is “cases per handler per month,” not “calls per agent per hour.” The customer experience metric is settlement time and outcome, not service level.
The two halves are connected (today’s FNOL volume becomes next month’s handling workload) but planned differently. Most operations get this right for FNOL and wrong for handling because the planning techniques are different.
Forecasting FNOL
FNOL volume has three layers.
Steady-state baseline. The everyday flow of motor incidents, household problems, business-as-usual claims. Behaves like normal contact centre arrival — seasonal, day-of-week, intraday pattern. Forecast it with the usual time-series methods. See seasonality multipliers.
Known events. School holidays, bank holiday Mondays (motor claims spike), heavy frost overnight, named storms tracked 3–7 days out. These are forecastable — if not the timing of the event, then the response curve once it’s identified.
Catastrophe (CAT) events. Major weather, large fires, a flood event, a riot, a property collapse. These are not forecastable in advance, but they have predictable response shapes once they happen. The forecast in CAT is less “what will volume be” and more “what curve does volume follow once an event triggers.”
The strongest FNOL forecasting teams maintain a CAT response library — a small archive of past events with the volume curve and duration. Storm Eunice 2022, a 1-in-30-year flood, the 2018 freeze — each has a characteristic shape. When a new event triggers, the forecaster matches it to the closest analogue in the library and produces a credible curve within hours. That curve is what scheduling and surge plans run from.
Forecasting handling capacity
Handling capacity is forecast in cases per handler per period — usually per month. The unit:
Cases handled = active handlers × cases per handler per month × (1 − non-productive time)
The variables that move it: complexity mix (a household subsidence case takes 10x the handler hours of a windscreen claim), regulatory burden (some lines of business have more compliance steps than others), system efficiency, handler experience curve, and the proportion of handler time spent on internal liaison rather than the customer.
Senior planners track cases per FTE per month as their headline capacity metric and treat it as something to be measured weekly and improved through process, not just staffed at. A 10% improvement in cases-per-FTE has the same effect as adding 10% to headcount — far cheaper and far more sustainable.
Skill boundaries are real
Service contact centres can usually cross-skill broadly. Claims operations can’t. Handler authority is tied to licensing, regulatory training, line-of-business specialism, and (in some markets) financial-services qualifications. A motor handler cannot pick up household claims without retraining. A junior handler cannot make settlement decisions above their authority limit without referral. A complex claims handler cannot be used for FNOL fill-in without losing their case continuity.
The planning implication: the “total FTE” number is misleading. The right view is FTE by skill, by line of business, by authority level, by licensed product. Capacity is a matrix, not a total. Bottlenecks usually show up in one cell of that matrix while the headline FTE number looks healthy.
Catastrophe planning — the discipline that separates serious claims operations
Major weather events test claims operations like nothing else. Volume can jump 8x to 20x in a single day. Standard operating models break inside hours. The operations that handle CAT events well do so because they have planned for them — before the event, not during.
The CAT response plan. A documented, owned, rehearsed plan that includes: trigger criteria (when does “normal” become CAT), decision authority (who calls the surge), staffing options (overtime, loan-in adjusters from other regions, contingent CAT teams from external partners, BPO surge cover), digital deflection (CAT-specific online claim forms, FAQ updates), and customer communications (claim acknowledgement SLA, expectation-setting messaging). Most insurers have a plan; the question is whether it gets rehearsed.
The CAT surge supply. Sustainable surge capacity isn’t achievable from the in-house team alone. The operations that handle big events well have pre-agreed contingent capacity: outsourced surge partners on retainers with clear trigger conditions, agency adjuster networks with framework agreements, regional loan-in agreements with sister operations. Negotiating surge capacity at the time of the event is a losing position; pre-negotiated capacity at fixed retainers is the operating model.
The CAT learning loop. After every meaningful event, run a structured review: trigger time vs response time, peak volume vs forecast volume, customer experience metrics during the event, handler wellbeing during and after, what the surge partner delivered. Write it down. The next event is six months away; the lessons from this one have to be captured while they’re fresh. The best claims operations have a CAT library that grows every year.
The regulatory dimension
Claims is one of the most heavily watched parts of a regulated insurer. In the UK, FCA Consumer Duty (see what Consumer Duty means for planners) sets explicit expectations on claims outcomes, fairness, vulnerable customer handling, and complaint resolution. In Ireland, the Central Bank’s Consumer Protection Code does similar work. The planning function isn’t directly accountable to the regulator — the legal entity is — but planning decisions shape the outcomes the regulator looks at.
Three practical implications.
Outcome metrics matter more than activity metrics. Settlement times, fair-outcome rates, complaint volumes, and vulnerable-customer handling matter more to the regulator than handler productivity. The MI pack and the planning rhythm should foreground outcomes alongside activity.
Vulnerable customers must be visibly served. See planning for vulnerable customers. Claims is the channel where vulnerability shows up most often; the operating model has to identify, route, and handle vulnerable customers appropriately.
Complaint capacity is planning capacity. Complaints are an output of how claims are handled. Plan for complaint volume as a percentage of claims volume; staff the complaint team accordingly. Operations that let the complaint queue build for two months attract regulatory attention.
The leadership cadences for a claims operation
Five cadences that keep a claims operation honest.
Daily FNOL stand-up. 15 minutes. Yesterday’s volume vs forecast, today’s outlook, any in-day risks. Identical to service operation rhythm.
Weekly handling review. Cases opened, closed, in flight by line of business and complexity. Ageing distribution. Customer-experience signal (NPS, complaint volume, regulatory escalations). The discipline of weekly review prevents the “quietly building backlog” failure that defines most under-managed claims operations.
Monthly capacity review. Cases-per-FTE trend, hiring pipeline, training pipeline, skill matrix coverage. Six-month outlook for capacity vs forecast.
Quarterly CAT preparedness review. Surge partners reaffirmed. Response plan walked through. Library updated. Communications templates refreshed. The discipline of not letting CAT preparation drift between events.
Annual operating-model review. Skill boundaries, authority levels, productivity assumptions, regulatory landscape, technology roadmap, customer outcome trend. The structural review that updates the operating model rather than letting it accrete.
Conclusion
Claims is not a service operation with longer handle times. It’s a different planning problem: two halves (FNOL and handling) with different units of capacity, a skill matrix that constrains substitution, a surge dimension that no normal capacity model accommodates, and a regulator paying close attention to outcomes. The teams that plan claims well treat each of those dimensions explicitly: separate FNOL and handling models, cases-per-FTE as the headline capacity metric, a documented and rehearsed CAT response, outcome-led MI, and a clear leadership cadence. The teams that don’t are usually the ones explaining a surge they didn’t see coming.
Pair this with what Consumer Duty means for planners, planning for vulnerable customers, planning with an outsourcer, and seasonality multipliers.